Re: [racket-dev] overwrite
On Tue, Mar 8, 2011 at 3:59 PM, John Clements cleme...@brinckerhoff.org wrote: On Fri, Mar 4, 2011 at 3:24 PM, Jakub Piotr Cłapa jpc...@zenburn.net wrote: On 27.02.11 23:33, Eli Barzilay wrote: Two hours ago, John Clements wrote: Add'l data point: I tried messing up the clock manually, by turning off NTP, but I was unable to duplicate the bug this way. NTP is unlikely to be the problem, since the times are saved on the filesystem, independently of the system time (or clock). And since Robby says that DrR compares only the time value regardless of being in the future, then there shouldn't be any problem with that. If NTP or clock drift was the reason then makefiles would not work and vim would also complain a lot (an this is not the case, at least for me). I do not use network mounts or git and I too get this message almost all the time on OS X. I will try to track it down using the patches posted in this thread. I got my first hit using these patches. I tried to save the file, and was informed that it had changed on disk since my last save from DrRacket, and did I want to overwrite it. Here's what I got last-saved-file-time: 1299611826 file-modified-time: 1299611832 These times are both quite close to when I tried to save. In retrospect, I should have added the output of (current-seconds) to the log output. I don't recall a six-second delay while trying to save. Hm. Yeah, I'm not seeing any clues in those numbers. How about also putting the filename into the log (the one that's being passed to file-or-directory-modify-seconds) and recording if it is the after-load-file or the after-save-file method that's updating the last-saved-file-time field? Robby _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev
Re: [racket-dev] overwrite
On 27.02.11 23:33, Eli Barzilay wrote: Two hours ago, John Clements wrote: Add'l data point: I tried messing up the clock manually, by turning off NTP, but I was unable to duplicate the bug this way. NTP is unlikely to be the problem, since the times are saved on the filesystem, independently of the system time (or clock). And since Robby says that DrR compares only the time value regardless of being in the future, then there shouldn't be any problem with that. If NTP or clock drift was the reason then makefiles would not work and vim would also complain a lot (an this is not the case, at least for me). I do not use network mounts or git and I too get this message almost all the time on OS X. I will try to track it down using the patches posted in this thread. -- regards, Jakub Piotr Cłapa _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev
Re: [racket-dev] overwrite
Yes, that's it. Also, put something into the after-save-file-method: diff --git a/collects/framework/private/editor.rkt b/collects/framework/private/editor.rkt index 0c8981b..cb5504a 100644 --- a/collects/framework/private/editor.rkt +++ b/collects/framework/private/editor.rkt @@ -190,7 +190,8 @@ (set! last-saved-file-time (and filename (file-exists? filename) - (file-or-directory-modify-seconds filename) + (file-or-directory-modify-seconds filename)) +(log-warning (format saved ~a at ~a filename last-saved-file-time) (inner (void) after-save-file success?)) and then when you see this strange behavior, maybe it the particular numbers that you're seeing might shed some light on which part of this is going wrong. Thanks! Robby On Sun, Feb 27, 2011 at 12:08 PM, John Clements cleme...@brinckerhoff.org wrote: On Feb 26, 2011, at 9:26 AM, Kathy Gray wrote: Same answers. -Kathy +1 I've seen this for ~ 6 months; I believe I've reported it before, but never followed up on it adequately. At least one of my students has also seen it. To add a few details: It usually happens on a run, asking if I want to revert the file. The message occurs on every subsequent run, until I save. When I save, DrRacket indicates that the file has changed on disk, and do I want to overwrite (yes). Then things are fine again. Is this the kind of annotation you were thinking of? oiseau:~/plt/collects/framework clements$ git diff private/editor.rkt diff --git a/collects/framework/private/editor.rkt b/collects/framework/private/editor.rkt index 0c8981b..9f21ce6 100644 --- a/collects/framework/private/editor.rkt +++ b/collects/framework/private/editor.rkt @@ -209,6 +209,9 @@ (and fn (file-exists? fn) (let ([ms (file-or-directory-modify-seconds fn)]) + (when ( last-saved-file-time ms) + (log-warning (format last-saved-file-time: ~v last-saved-file-time)) + (log-warning (format file-modified-time: ~v ms))) ( last-saved-file-time ms)) (define has-focus #f) oiseau:~/plt/collects/framework clements$ Looking at this code, I'm betting on Neil's suggestion of NTP, though I realize that the mechanism isn't obvious... perhaps the OS adjusts the save times of recently-saved files? John Clements On 26 Feb 2011, at 5:23:56, Matthias Felleisen wrote: On Feb 26, 2011, at 12:12 PM, Robby Findler wrote: Are any of you, by any chance: - using a teaching language (via the language menu, not #lang 2htdp/bsl or similar) #lang racket - hitting run and then save right *after* run no when you get this message? Robby On Sat, Feb 26, 2011 at 11:10 AM, Nadeem Abdul Hamid nad...@acm.org wrote: I get this a lot, using just the binary installer version downloaded from racket-lang.org. And it happens even when I'm not using check-syntax, though it might happen more frequently when I do use check-syntax; hard to tell. On Sat, Feb 26, 2011 at 12:05 PM, Matthias Felleisen matth...@ccs.neu.edu wrote: Well it just occurred for a file under Git but WITHOUT check syntax. On Feb 26, 2011, at 11:48 AM, Matthias Felleisen wrote: I confirm the check syntax observation. On Feb 26, 2011, at 11:39 AM, Kathy Gray wrote: I also see this quite frequently (for files not under Git control). I was thinking at one point it might be related to when I run check syntax, but I haven't confirmed a repeatable pattern. -Kathy On 26 Feb 2011, at 4:36:37, Matthias Felleisen wrote: The files I have been editing this morning are not under Git control. (And yes, I have on one occasion checked the file via Emacs and didn't see any difference.) On Feb 26, 2011, at 11:35 AM, Jay McCarthy wrote: It's from git touching the files and giving them a new timestamp. It probably didn't really change. Jay 2011/2/26 Matthias Felleisen matth...@ccs.neu.edu: When I use drracket, I frequently get a warning that my file has been modified on disk and the question of whether I want to save the file or revert. Is anyone else suffering from this problem? _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev -- Jay McCarthy j...@cs.byu.edu Assistant Professor / Brigham Young University http://faculty.cs.byu.edu/~jay The glory of God is Intelligence - DC 93 _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev _ For list-related administrative tasks:
Re: [racket-dev] overwrite
On Feb 27, 2011, at 11:42 AM, Robby Findler wrote: Yes, that's it. Also, put something into the after-save-file-method: diff --git a/collects/framework/private/editor.rkt b/collects/framework/private/editor.rkt index 0c8981b..cb5504a 100644 --- a/collects/framework/private/editor.rkt +++ b/collects/framework/private/editor.rkt @@ -190,7 +190,8 @@ (set! last-saved-file-time (and filename (file-exists? filename) - (file-or-directory-modify-seconds filename) + (file-or-directory-modify-seconds filename)) +(log-warning (format saved ~a at ~a filename last-saved-file-time) (inner (void) after-save-file success?)) and then when you see this strange behavior, maybe it the particular numbers that you're seeing might shed some light on which part of this is going wrong. Add'l data point: I tried messing up the clock manually, by turning off NTP, but I was unable to duplicate the bug this way. John smime.p7s Description: S/MIME cryptographic signature _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev
Re: [racket-dev] overwrite
Looking at this code, I'm betting on Neil's suggestion of NTP, though I realize that the mechanism isn't obvious... perhaps the OS adjusts the save times of recently-saved files? I haven't followed this discussion closely, nor looked at the relevant code. However :) From what I know about NTP, it will never make time go backwards, so I doubt it's the culprit. (When NTP determines that your clock is later than it should be, it merely slows it down a bit until the true time catches up with it.) _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev
Re: [racket-dev] overwrite
Two hours ago, John Clements wrote: Add'l data point: I tried messing up the clock manually, by turning off NTP, but I was unable to duplicate the bug this way. NTP is unlikely to be the problem, since the times are saved on the filesystem, independently of the system time (or clock). And since Robby says that DrR compares only the time value regardless of being in the future, then there shouldn't be any problem with that. (It might, however, lead to problems if there's something that can change the file's time when it's in the future -- for example, if compiling the file tries to guarantee that it is not in the future for the compiled file to be further in time, and if DrR causes this by the automatic compilation. (This is not a guess about the problem, just an example of how it could break.)) As for git, it shouldn't lead to times in the future, since unlike other systems (like svn), it does not have a time stamp for each file. Instead, the idea is that it only changes files that it touches and uses the current time. This is why a rebase operation can lead to a lot of time changes -- since it will undo your commit(s), then fast-forward the repository, then replay the commits, so you end up with many files that have the same contents, but a different time. (And as an unrelated reminder -- if this bothers anyone, I have a wrapper that properly restores the time for such operations.) -- ((lambda (x) (x x)) (lambda (x) (x x))) Eli Barzilay: http://barzilay.org/ Maze is Life! _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev
Re: [racket-dev] overwrite
The files I have been editing this morning are not under Git control. (And yes, I have on one occasion checked the file via Emacs and didn't see any difference.) On Feb 26, 2011, at 11:35 AM, Jay McCarthy wrote: It's from git touching the files and giving them a new timestamp. It probably didn't really change. Jay 2011/2/26 Matthias Felleisen matth...@ccs.neu.edu: When I use drracket, I frequently get a warning that my file has been modified on disk and the question of whether I want to save the file or revert. Is anyone else suffering from this problem? _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev -- Jay McCarthy j...@cs.byu.edu Assistant Professor / Brigham Young University http://faculty.cs.byu.edu/~jay The glory of God is Intelligence - DC 93 _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev
Re: [racket-dev] overwrite
I confirm the check syntax observation. On Feb 26, 2011, at 11:39 AM, Kathy Gray wrote: I also see this quite frequently (for files not under Git control). I was thinking at one point it might be related to when I run check syntax, but I haven't confirmed a repeatable pattern. -Kathy On 26 Feb 2011, at 4:36:37, Matthias Felleisen wrote: The files I have been editing this morning are not under Git control. (And yes, I have on one occasion checked the file via Emacs and didn't see any difference.) On Feb 26, 2011, at 11:35 AM, Jay McCarthy wrote: It's from git touching the files and giving them a new timestamp. It probably didn't really change. Jay 2011/2/26 Matthias Felleisen matth...@ccs.neu.edu: When I use drracket, I frequently get a warning that my file has been modified on disk and the question of whether I want to save the file or revert. Is anyone else suffering from this problem? _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev -- Jay McCarthy j...@cs.byu.edu Assistant Professor / Brigham Young University http://faculty.cs.byu.edu/~jay The glory of God is Intelligence - DC 93 _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev
Re: [racket-dev] overwrite
Well it just occurred for a file under Git but WITHOUT check syntax. On Feb 26, 2011, at 11:48 AM, Matthias Felleisen wrote: I confirm the check syntax observation. On Feb 26, 2011, at 11:39 AM, Kathy Gray wrote: I also see this quite frequently (for files not under Git control). I was thinking at one point it might be related to when I run check syntax, but I haven't confirmed a repeatable pattern. -Kathy On 26 Feb 2011, at 4:36:37, Matthias Felleisen wrote: The files I have been editing this morning are not under Git control. (And yes, I have on one occasion checked the file via Emacs and didn't see any difference.) On Feb 26, 2011, at 11:35 AM, Jay McCarthy wrote: It's from git touching the files and giving them a new timestamp. It probably didn't really change. Jay 2011/2/26 Matthias Felleisen matth...@ccs.neu.edu: When I use drracket, I frequently get a warning that my file has been modified on disk and the question of whether I want to save the file or revert. Is anyone else suffering from this problem? _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev -- Jay McCarthy j...@cs.byu.edu Assistant Professor / Brigham Young University http://faculty.cs.byu.edu/~jay The glory of God is Intelligence - DC 93 _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev
Re: [racket-dev] overwrite
I get this a lot, using just the binary installer version downloaded from racket-lang.org. And it happens even when I'm not using check-syntax, though it might happen more frequently when I do use check-syntax; hard to tell. On Sat, Feb 26, 2011 at 12:05 PM, Matthias Felleisen matth...@ccs.neu.edu wrote: Well it just occurred for a file under Git but WITHOUT check syntax. On Feb 26, 2011, at 11:48 AM, Matthias Felleisen wrote: I confirm the check syntax observation. On Feb 26, 2011, at 11:39 AM, Kathy Gray wrote: I also see this quite frequently (for files not under Git control). I was thinking at one point it might be related to when I run check syntax, but I haven't confirmed a repeatable pattern. -Kathy On 26 Feb 2011, at 4:36:37, Matthias Felleisen wrote: The files I have been editing this morning are not under Git control. (And yes, I have on one occasion checked the file via Emacs and didn't see any difference.) On Feb 26, 2011, at 11:35 AM, Jay McCarthy wrote: It's from git touching the files and giving them a new timestamp. It probably didn't really change. Jay 2011/2/26 Matthias Felleisen matth...@ccs.neu.edu: When I use drracket, I frequently get a warning that my file has been modified on disk and the question of whether I want to save the file or revert. Is anyone else suffering from this problem? _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev -- Jay McCarthy j...@cs.byu.edu Assistant Professor / Brigham Young University http://faculty.cs.byu.edu/~jay The glory of God is Intelligence - DC 93 _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev -- Nadeem Abdul Hamid Associate Professor, Computer Science Berry College PO Box 5014 2277 Martha Berry Hwy NW Mount Berry, GA 30149-5014 (706) 368-5632 http://cs.berry.edu/~nhamid/ _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev
Re: [racket-dev] overwrite
Are any of you, by any chance: - using a teaching language (via the language menu, not #lang 2htdp/bsl or similar) - hitting run and then save right *after* run when you get this message? Robby On Sat, Feb 26, 2011 at 11:10 AM, Nadeem Abdul Hamid nad...@acm.org wrote: I get this a lot, using just the binary installer version downloaded from racket-lang.org. And it happens even when I'm not using check-syntax, though it might happen more frequently when I do use check-syntax; hard to tell. On Sat, Feb 26, 2011 at 12:05 PM, Matthias Felleisen matth...@ccs.neu.edu wrote: Well it just occurred for a file under Git but WITHOUT check syntax. On Feb 26, 2011, at 11:48 AM, Matthias Felleisen wrote: I confirm the check syntax observation. On Feb 26, 2011, at 11:39 AM, Kathy Gray wrote: I also see this quite frequently (for files not under Git control). I was thinking at one point it might be related to when I run check syntax, but I haven't confirmed a repeatable pattern. -Kathy On 26 Feb 2011, at 4:36:37, Matthias Felleisen wrote: The files I have been editing this morning are not under Git control. (And yes, I have on one occasion checked the file via Emacs and didn't see any difference.) On Feb 26, 2011, at 11:35 AM, Jay McCarthy wrote: It's from git touching the files and giving them a new timestamp. It probably didn't really change. Jay 2011/2/26 Matthias Felleisen matth...@ccs.neu.edu: When I use drracket, I frequently get a warning that my file has been modified on disk and the question of whether I want to save the file or revert. Is anyone else suffering from this problem? _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev -- Jay McCarthy j...@cs.byu.edu Assistant Professor / Brigham Young University http://faculty.cs.byu.edu/~jay The glory of God is Intelligence - DC 93 _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev -- Nadeem Abdul Hamid Associate Professor, Computer Science Berry College PO Box 5014 2277 Martha Berry Hwy NW Mount Berry, GA 30149-5014 (706) 368-5632 http://cs.berry.edu/~nhamid/ _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev
Re: [racket-dev] overwrite
Same answers. -Kathy On 26 Feb 2011, at 5:23:56, Matthias Felleisen wrote: On Feb 26, 2011, at 12:12 PM, Robby Findler wrote: Are any of you, by any chance: - using a teaching language (via the language menu, not #lang 2htdp/bsl or similar) #lang racket - hitting run and then save right *after* run no when you get this message? Robby On Sat, Feb 26, 2011 at 11:10 AM, Nadeem Abdul Hamid nad...@acm.org wrote: I get this a lot, using just the binary installer version downloaded from racket-lang.org. And it happens even when I'm not using check-syntax, though it might happen more frequently when I do use check-syntax; hard to tell. On Sat, Feb 26, 2011 at 12:05 PM, Matthias Felleisen matth...@ccs.neu.edu wrote: Well it just occurred for a file under Git but WITHOUT check syntax. On Feb 26, 2011, at 11:48 AM, Matthias Felleisen wrote: I confirm the check syntax observation. On Feb 26, 2011, at 11:39 AM, Kathy Gray wrote: I also see this quite frequently (for files not under Git control). I was thinking at one point it might be related to when I run check syntax, but I haven't confirmed a repeatable pattern. -Kathy On 26 Feb 2011, at 4:36:37, Matthias Felleisen wrote: The files I have been editing this morning are not under Git control. (And yes, I have on one occasion checked the file via Emacs and didn't see any difference.) On Feb 26, 2011, at 11:35 AM, Jay McCarthy wrote: It's from git touching the files and giving them a new timestamp. It probably didn't really change. Jay 2011/2/26 Matthias Felleisen matth...@ccs.neu.edu: When I use drracket, I frequently get a warning that my file has been modified on disk and the question of whether I want to save the file or revert. Is anyone else suffering from this problem? _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev -- Jay McCarthy j...@cs.byu.edu Assistant Professor / Brigham Young University http://faculty.cs.byu.edu/~jay The glory of God is Intelligence - DC 93 _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev -- Nadeem Abdul Hamid Associate Professor, Computer Science Berry College PO Box 5014 2277 Martha Berry Hwy NW Mount Berry, GA 30149-5014 (706) 368-5632 http://cs.berry.edu/~nhamid/ _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev
Re: [racket-dev] overwrite
The code compares the date of the file when it was last saved to the date of the file when run is clicked. Robby On Sat, Feb 26, 2011 at 12:50 PM, Neil Van Dyke n...@neilvandyke.org wrote: Dunno whether this helps... I haven't looked at the pertinent Racket code, but errors like this often happen in Unix applications because the mtime of the file is in the future relative to the current time clock. Less commonly, because the mtime changed since the file was read by the app or since the app last checked. The most common case can happen because a VC/CM system is updating the times (someone mentioned Git, and some other systems do this), because it's a network filesystem and the server and workstation clocks aren't synced sufficiently, or because NTP or some other mechanism dangerously jumped the current time clock in reverse (usually a good NTP daemon will slow the clock rather than jump in reverse). Robby Findler wrote at 02/26/2011 01:39 PM: Looking at the code, I don't see anything suspicious (assuming that after-save-file and file-or-directory-modify-seconds work properly, that is). -- http://www.neilvandyke.org/ _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev _ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/dev
